perplexity-performance-tuning

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The skill accepts an outsider-provided query at runtime (e.g., from the triggering user phrase) and sends it directly as messages: [{role:"user", content: query}] to Perplexity at perplexity.chat.completions.create, so the LLM ingests free text without selecting a specific trusted item first.